Here's a check list for diagnosing unknown hardware faults such as sudden freezing, etc:
Check for obvious visual clues outside and inside the box:
Try removing all inessential peripherals such as USB devices, etc.
Is excessive dust build-up clogging heatsinks, ventilation, floppy or CD drives, etc?
If so, being sure beforehand to remove things like the cover screws out of harm's way,
this can be carefully removed with the nozzle attachment of an ordinary domestic
vacuum cleaner, holding the nozzle firmly at both the handle and the business end to
prevent the suction causing sudden violent movement.
Are all the fans running -
CPU, chipset if any,
PSU, system, case?
Do any components show signs of heat damage?
Are all cables / plugin boards mounted securely - Memory,
AGP,
PCI(X),
ISA,
(E)IDE,
SCSI,
USB?
If the problem seems to be hard disk (HD) related:
Run ChkDsk (Windows) or fsck (Linux) or other disk
diagnostics program to check for file system errors, and to scan the disk surfaces for
defects.
If you swap cables around, does the problem move with a particular cable?
If you swap the motherboard controllers that the HDs are connected to, is the problem
associated with a particular controller? If so, disable it in the BIOS and connect
the HDs to another.
Can the PSU cope with demand? A symptom of inadequate PSU can be that when a HD
reaches the power-saving idle timeout and powers down, when it is next accessed, it takes
excessively long to power up. The system may even cycle between briefly allowing a
few keystrokes or mouse-clicks and freezing for up to a minute. If you power down
and disconnect enough inessential hardware such as floppy/CD drives or non-system HDs, such
symptoms disappear.
If the problem seems to be BIOS related (the BIOS
can usually be entered from system power up by pressing and holding down a particular key, often
displayed on the screen at the appropriate time to press it, usually <Delete>
or one of the <F?> 'function' keys):
Check the diagnostics in the BIOS - are all the temperatures and voltages ok?
Are the BIOS settings being lost? This used to be quite a common problem, but seems
rarer now. During power disconnection, they are maintained (in low-current-drain
memory called CMOS) by a
battery - usually a CR2032 lithium watch battery. If this is running low,
BIOS settings may become corrupt on power disconnection, including CPU / chipset
clock speed, system time and date, boot device order, enabled/disabled state of integrated
devices such as IDE controllers, etc. Check the BIOS settings are as expected after
power loss. If not, change the battery.
Have the motherboard manufacturers released a newer version of the BIOS to cure your
problem? Check their website and if necessary follow their instructions to upgrade
the BIOS.
Run MemTest to test system memory - Linux & some recent versions of
Windows sometimes include it as a 'safe' mode or 'Recovery Console' boot option, otherwise download the
floppy image, make up the floppy, and boot from it:
http://www.memtest86.com/download.html
Can you boot from, say, a Windows or Linux installation CD, and/or run a Linux Live CD? If so,
perhaps the problem is Operating System (OS) or software
rather than hardware related.
If your OS is Windows, are there any clues in the System Event Log?
Install any hardware monitor / diagnostics program that may have come with the PC or motherboard.
Choose settings that log as many potential problems as possible, and examine the log after each crash.
However, be aware that they tend to be unreliable. Most do not run as a service, which they
should, so won't run, say, while the PC is at the logon screen. Even more absurd, at least some
Gigabyte programs hibernate when minimised to the system tray! Therefore, keep the program window
open at all times until you've nailed the problem.
If still no solution, and you have spares, try swapping out hardware, one piece at a time.